Tubal ligation and incidence of 26 site-specific cancers in the Million Women Study.

The study population included 1 278 783 women without prior cancer, average age 56.1 (s.d. 4.9) at recruitment, 23% of whom reported a tubal ligation, at an average age of 34.5 (s.d. 5.3). Table 1 shows baseline characteristics of the cohort. A total of 167 430 cancers accrued, after a mean follow-up of 13.8 years (s.d. 3.4). The average age at cancer diagnosis was 65.6 years (s.d. 6.5). There was no association between tubal ligation and risk of all cancers combined (RR=1.00, 95% confidence interval (CI): 0.98–1.01). Figure 1 displays adjusted RRs and 95% CIs of 26 site- or type-specific cancers among women with vs without tubal ligation. There was a 20% reduction for ovarian cancer risk (RR=0.80, 95% CI: 0.76–0.85; P <0.001) and a reduction of similar magnitude for peritoneal cancers (RR=0.81, 95% CI: 0.66–0.98; P =0.03) (ICD-10 code C48, which includes both peritoneal and retroperitoneal cancers). We found a large reduction in risk of cancers of the fallopian tube, but confidence intervals were wide (RR: 0.60, 95% CI: 0.37–0.96, P =0.04). We also found increased risk of anal cancer (RR=1.34, 95% CI: 1.11–1.63; P =0.002), as previously reported ( Coffey et al , 2015 ). The small increase in risk of lung cancer (RR=1.09, 95% CI: 1.05–1.13) may well be due to residual confounding from smoking — as women with tubal ligation were much more likely to be current smokers, and to smoke more cigarettes per day, than women without tubal ligation. There was no association between tubal ligation and lung cancer risk among never-smokers (RR=1.04; 95% CI: 0.91–1.19). There were no significant associations between tubal ligation and risk of cancers at the other sites, including cancers of the endometrium, cervix, breast and colorectum.

The Million Women Study is a prospective study of 1.3 million UK women, recruited in 1996–2001. Participants completed a questionnaire at recruitment, on socio-demographic, reproductive and lifestyle factors. Information on tubal ligation came from responses to the question ‘Have you been sterilised (had your tubes tied)?' on the recruitment questionnaire. The study design and methods have been described previously ( The Million Women Study Collaborative Group, 1999 ; Million Women Study Collaborators, 2003 ) and questionnaires can be viewed online at http://www.millionwomenstudy.org . Follow-up for outcomes was based on routine registers: participants have been flagged on the NHS Central Register for cancer registrations and deaths. Information on cancer provided to investigators includes the date of registration and cancer site (coded using the 10th revision of the International Classification of Diseases, ICD-10) ( World Health Organization, 1992 ). All participants gave written consent to follow-up at recruitment. Ethical approval was granted by the Oxford and Anglia Multi-Centre Research Ethics Committee (MREC 97/01). Women were excluded from the analyses if they had invasive cancer other than non-melanoma skin cancer (ICD-10 code C44) before recruitment ( n =66 221 (5%)), or missing data on tubal ligation ( n =40 612 (3%)). The remaining women ( n =1 278 783) contributed person-years from the date of recruitment until the earliest of the date of registration of any cancer (other than non-melanoma skin cancer), the date of death, or last date of follow-up (31st December 2013). About 1% of participants were lost to follow-up and contributed person-years until the date they were lost. Cox (proportional hazards) regression models were used to estimate hazard ratios (referred to as relative risks (RRs)) of developing a given cancer by tubal ligation status. Attained age was the underlying time variable. Analyses were run separately for each of 26 cancer types or sites, defined by ICD-10 code, and for all cancer (any invasive cancer other than non-melanoma skin cancer, ICD-10 C44). Analyses were limited to cancer types for which 500 or more cases with an identified primary site had accrued. In addition, we conducted an analysis for fallopian tube cancers (ICD-10 C57.0), despite having fewer than 500 cases, as there is a reasonable prior hypothesis that surgery to the fallopian tube might affect the incidence of cancer at this site. All analyses were stratified by geographical region (10 regions corresponding to the areas covered by the cancer registries) and quintiles of socioeconomic status (based on the Townsend deprivation index) ( Townsend et al , 1988 ), and additionally adjusted for parity (0, 1, 2, 3+), age at first birth (<20, 20–24, 25–29, 30+), smoking (never, past, current 7 & ⩽14 units, >14 units), frequency of strenuous exercise (<once/week, ⩾once/week), use of menopausal hormones (never, ever), use of the oral contraceptive pill (never, ever), hysterectomy (no, yes), and body mass index (<25 kg m −2 , 25–29  kg m −2 , 30+  kg m −2 ). All adjustment variables were as reported at recruitment. For adjustment and stratification variables, missing values were assigned to a separate category. Exposure information was either missing or reported as unknown for ⩽6% of women for all potential confounders. Analyses of ovarian cancer risk were restricted to women without bilateral oophorectomy; those of endometrial and cervical cancer were restricted to women without hysterectomy; those of fallopian tube cancer were restricted to women without either bilateral oophorectomy or hysterectomy (as the fallopian tubes might potentially be removed as part of either surgery). Analyses were performed in Stata-14 ( StataCorp, 2015 ). Figures were drawn in R ( R Development Core Team, 2015 ) using Matthew Arnold's ‘Jasper' package ( Arnold, 2015 ).

To our knowledge, this is the largest and most comprehensive analysis so far of the association between tubal ligation and subsequent risk of cancer. We found that tubal ligation was significantly associated with reduced risks of ovarian, peritoneal, and fallopian tube cancers, and an increased risk of anal cancer, but was not associated with breast, endometrial or cervical cancer, as had been suggested by others. The reduction in ovarian cancer risk has been reported by others ( Cibula et al , 2011 ; Rice et al , 2012 ; Sieh et al , 2013 ). We previously reported significant heterogeneity by histological type of ovarian cancer, finding that tubal ligation was associated with a modest reduction of serous tumours overall (RR=0.84, 95% CI: 0.77–0.92), and about a 20% reduction in risk of high-grade serous carcinoma, and almost a halving of risk of endometrioid and clear cell ovarian tumours ( Gaitskell et al , 2016 ). These findings are consistent with hypotheses regarding possible different origins of the ovarian cancer histotypes, and that many cases of high-grade serous ovarian cancer may arise from precursor lesions within the fallopian tubes, while some endometrioid and clear cell tumours may develop from endometriosis ( Kurman and Shih, 2011 ; Prat, 2012 ). Although the precise mechanism by which tubal ligation reduces the risk of ovarian cancer is unclear, part of the explanation could be that it acts as a physical barrier to pro-cancerous substances passing through the fallopian tubes to the ovaries (whether endometriosis, fallopian tube epithelial cells, infectious agents, or exogenous carcinogens). Few other studies have reported on the association between tubal ligation and risk of peritoneal cancer, and they have tended to be limited by small numbers of cases. Our study included 730 cases of peritoneal cancer; two retrospective studies have reported their findings—one study including 62 cases found no association ( Grant et al , 2010 ); and another including 22 cases reported a reduced risk ( Purdie et al , 1995 ; Green et al , 1997 ). We believe that this is the first study to report that tubal ligation is associated with a significant reduction in risk of fallopian tube cancer ( Riska et al , 2007 ; Vicus et al , 2010 ). Our finding that tubal ligation is associated with a reduction in risk of ovarian, peritoneal and fallopian tube cancers perhaps reflects their histological and clinical similarity, and possible similar site of origin ( Kurman et al , 2014 ). The majority of the peritoneal and fallopian tube cancers with specified histology were serous carcinomas, which is also the most common histotype of ovarian cancer ( Gaitskell et al , 2016 ). There is increasing evidence from histological, molecular and genetics studies that many high-grade serous carcinomas found in the ovary or peritoneum may have originated from precursor lesions in the fallopian tube (reviewed in ( Nik et al , 2014 ; Perets and Drapkin, 2016 )). As the majority of the putative precursor lesions for high-grade serous cancers are found within the fimbrial end of the fallopian tube ( Przybycin et al , 2010 ; Gilks et al , 2015 ), adjacent to the ovary, it is perhaps surprising that tubal ligation (which normally involves surgery to the mid-portion of the tube) should reduce the risk of such tumours. It may be that, historically, some surgical sterilisation procedures involved cutting and removing a portion of the fallopian tube, rather than simply clipping the tube (as is now more common). Alternatively, it may be that surgery causes local vascular changes with consequent effects on the tubal epithelium. One of the limitations of our study is that we do not have information on the type of surgical procedure performed, and thus are unable to explore possible differences in association between different surgical techniques. For endometrial cancer, a reduced risk associated with tubal ligation was reported in a Danish cohort ( Kjaer et al , 2004 ), but we and others have not replicated this association ( Castellsague et al , 1996 ; Lacey et al , 2000 ). One group of investigators reported that tubal ligation was associated with an increased risk of cervical cancer ( Mathews et al , 2012 ), but again, neither other findings ( Li and Thomas, 2000 ) nor ours replicated this. Reports of the relationship between tubal ligation and breast cancer risk have been varied, including suggestions of either a reduced ( Kreiger et al , 1999 ; Calle et al , 2001 ) or increased risk ( Irwin et al , 1988 ), although the majority of studies reported no significant association ( Brinton et al , 2000 ; Eliassen et al , 2006 ; Iversen et al , 2007 ; Dorjgochoo et al , 2009 ; Press et al , 2011 ; Gaudet et al , 2013 ; Nichols et al , 2013 ), consistent with our findings. We have previously reported in detail on the risk factors for anal cancer in the Million Women Study ( Coffey et al , 2015 ); factors associated with an increased risk of anal cancer included tubal ligation, nulliparity, smoking, past use of the oral contraceptive pill, and not living with a partner. The strongest association was with a history of cervical intraepithelial neoplasia grade 3 (RR=4.03, 95% CI: 2.59–6.28), reflecting the importance of high-risk strains of human papilloma virus as causative agents of both cervical ( Walboomers et al , 1999 ) and anal ( International Agency for Research on Cancer, 2009 ) cancers. It is unclear why tubal ligation should be associated with an increased risk of anal cancer, and it may be that this apparent association reflects confounding by sexual behaviour and exposure to human papilloma virus. For example, as tubal ligation is usually performed for permanent contraception, it is likely to be more common among women who are sexually active, and who thus may be more likely to be exposed to sexually-transmitted infections, such as human papilloma virus, and the consequent increased risk of anal cancer. There have been reports on possible relationships between tubal ligation and other cancers, including possible reductions in risk of colorectal ( Cape and Kreiger, 1999 ; Rosenblatt et al , 2004 ) and stomach cancer ( Dorjgochoo et al , 2009 ), and increased risks of thyroid cancer ( Braganza et al , 2014 ) and lymphatic and haematopoietic malignancies ( Kjaer et al , 2004 ). We did not replicate any of these findings. The large size of the cohort, the individual information on possible confounding factors, and the complete and long follow-up, provided reliable estimates of risks associated with tubal ligation for 26 specific cancer sites, even relatively uncommon ones. We found no association between tubal ligation and the risk of cancers of the endometrium, breast, or cervix. By contrast, tubal ligation is associated with a clear reduction in risk of ovarian cancer, a reduction of similar magnitude of peritoneal cancer, and a reduction of fallopian tube cancer. That tubal ligation is associated with a reduced risk of cancers of the ovary, peritoneum and fallopian tube, but not of other hormonally-related cancers, is consistent with the hypothesis that many of the cancers at these sites have a shared origin in the fallopian tube, and that tubal ligation reduces cancer risk by acting as a barrier to cells, carcinogens or other agents reaching the ovary and peritoneal cavity, rather than by affecting hormone levels.